Hacker News: jnwatson

New comment by jnwatson in "Tuxedo No. 2 – Cocktail recipes"

jnwatson — Sun, 09 Aug 2026 22:59:09 +0000

I like cocktails. I like cocktail recipes. I don't know why this is on hacker news thought.

New comment by jnwatson in "Show HN: Airy – Free, fast, and simple voice content creation"

jnwatson — Sun, 09 Aug 2026 16:12:23 +0000

All but one of the voices sounds tinny. The timing and cadence was good though.

Is this intended for anime autodubbing? There's only one voice (Rowan) that is remotely traditional broadcaster-style.

New comment by jnwatson in "Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD"

jnwatson — Sat, 08 Aug 2026 22:24:00 +0000

While the AGPL is untested, what a derived work is is less so. It is entirely outside the bounds of the intentions of the authors of the AGPL for clients to be infected.

New comment by jnwatson in "Radical Study Suggests Life on Earth Arose Twice"

jnwatson — Fri, 07 Aug 2026 17:18:00 +0000

The article does make it clear, though the headline misleads.

New comment by jnwatson in "Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD"

jnwatson — Fri, 07 Aug 2026 15:59:11 +0000

Why would AGPL force you to open source your app? Unless you literally compile your app with pgrust by modifying the pgrust source code, you're safe. Clients aren't bound by the AGPL because they aren't derived works.

New comment by jnwatson in "Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD"

jnwatson — Fri, 07 Aug 2026 15:56:18 +0000

The floating point comparison bug is nightmare fuel. I could look at that for years and never spot the mistake.

New comment by jnwatson in "Almost no skill required to cook a steak"

jnwatson — Thu, 06 Aug 2026 16:54:05 +0000

It is a great analogy, as the vast majority of software tasks are CRUD or glue logic. At this point, relatively narrowly-defined tasks are a solved problem with frontier models. It is only more advanced tasks with broad scope and large context that frontier models have more limited results.

New comment by jnwatson in "I’m leaving OpenAI to build telepathy"

jnwatson — Wed, 05 Aug 2026 22:30:21 +0000

They are going to have to come up with better portable noninvasive technology.

Technically fMRI is noninvasive; it is just a bit inconvenient for everyday use.

New comment by jnwatson in "Why Erdős Problems Are Falling to AI"

jnwatson — Wed, 05 Aug 2026 14:39:37 +0000

Isn't most of math this way? Occasionally, we find a practical use for it, but that's usually not the point.

New comment by jnwatson in "AI Data Centers Are Driving Up Power Bills – This Map Shows Where"

jnwatson — Tue, 04 Aug 2026 19:43:13 +0000

Electricity production isn't a free market in the vast majority of municipalities. Market systems only work when demand and supply are relatively unconstrained. In this case, supply is strongly constrained.

New comment by jnwatson in "Safe Lock-free Primitives with iceoryx2's ByteAtomic"

jnwatson — Tue, 04 Aug 2026 13:58:07 +0000

Also, there's no guarantee of progress. The writer can starve the reader forever.

New comment by jnwatson in "My personal AI benchmark: “Generate an SVG of a frog with a Habsburg jaw”"

jnwatson — Mon, 03 Aug 2026 02:02:02 +0000

Fable 5 on Max knocks it out of the park:

https://imgur.com/a/usR8K7G

Definitely has some creative flourishes.

(I made no extra prompting. Just the above text. Single shot.)

New comment by jnwatson in "MkLinux and the pimped-out Apple Workgroup Server 9150"

jnwatson — Sun, 02 Aug 2026 14:44:49 +0000

Some teenagers saved up money for a car. I saved up money for a PowerMac 7100. It cost me $3k, with a solid student discount.

I used it throughout college and beyond. I upgraded the RAM, the L2 cache. I even added one of the Sonnet G3 cards the article shows.

During its latter years, it only ran MkLinux.

New comment by jnwatson in "Run Kimi K3 using 29 GB of RAM at 0.50 tok/s"

jnwatson — Sat, 01 Aug 2026 13:13:57 +0000

There was another project posted a few days ago that was quite similar and someone asked the same thing to the author. They said they tried that first and they got a 10x speed boost by doing it manually. It is the same reason why database engines write their own backing cache logic: the kernel's paging logic is generic and on-demand, while someone that's aware of the actual usage patterns can prefetch and pipeline the right data before it is used.

New comment by jnwatson in "Elevated errors on Claude Opus 5"

jnwatson — Mon, 27 Jul 2026 22:09:27 +0000

It seems to have trouble remembering the whole context, even when its limit is only half full. Three times this weekend I've had to switch to Fable, where I literally ask "review the recent conversation and tell me where we went offtrack" and Fable immediately identifies the problems that Opus was having.

I'm doing data science stuff so it isn't super complicated code; it is about applying valid statistical procedures and techniques. Still, on the code part, Opus 5 had a lot of trouble merging 2 branches yesterday...

On a tangent, I am beginning to understand why we have replication crisis in academia. I thought C++ was full of footguns; it has nothing on statistics. With statistics, you don't get a compiler error or a crash when you hold it wrong.

New comment by jnwatson in "Decker, a platform that builds on the legacy of Hypercard and classic macOS"

jnwatson — Mon, 27 Jul 2026 04:29:11 +0000

Suggested retail price for Filemaker when it was originally released in 1985 for Macintosh was $195. That's about $600 these days.

It is about as expensive as it ever was.

New comment by jnwatson in "Claude Opus 5"

jnwatson — Sat, 25 Jul 2026 13:51:44 +0000

Interestingly, Fable caught the gist and I didn't have to enumerate every Claudism. "Punchy" seemed to be the operative word. My new operating hypothesis is that Claude (Fable in particular) defaults to optimize for concision and "turn of phrase". If we can turn that off, the prose is way more natural.

Claude responded "The arguments and structure are unchanged, but sentences now state claims directly instead of building to a turn of phrase."

I'm OK with colons and semicolons as I tend to write that way.

New comment by jnwatson in "ARC-AGI Leaderboard"

jnwatson — Sat, 25 Jul 2026 13:05:59 +0000

I was just thinking they need to mark each model per benchmark as "model released before the benchmark was released" and "model released after the benchmark was released".

New comment by jnwatson in "Claude Opus 5"

jnwatson — Sat, 25 Jul 2026 04:51:29 +0000

I recently wrote a short paper with Fable, and, with some prodding, I was able to get some non-painful prose out of it.

I just found my prompt:

The writing style could really use some work. Avoid Claude-isms like "stated fairly", em dashes, "load-bearing", overly punchy phrasing like "keep the signal, govern the response". This is a technical document, not a marketing campaign.

New comment by jnwatson in "SpaceX at $100 Would Imply Zero AI Value, Morgan Stanley Says"

jnwatson — Fri, 24 Jul 2026 23:09:04 +0000

And it is likely going to happen a fourth time when SpaceX merges with Tesla. The first time was when Tesla bought Solar City.